Fix the ctgrind errors in the 25519 tests

[brian-pane]

No description provided.

[codspeed-hq]

Merging this PR will not alter performance

✅ 166 untouched benchmarks

---

_{Comparing brian-pane:valgrind-fixes (f206525) with main (f27b0a7)}

[codecov]

Codecov Report

❌ Patch coverage is 94.73684% with 1 line in your changes missing coverage. Please review.
✅ Project coverage is 99.76%. Comparing base (f27b0a7) to head (f206525).

Files with missing lines	Patch %	Lines
graviola/src/high/ed25519.rs	87.50%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #171      +/-   ##
==========================================
- Coverage   99.76%   99.76%   -0.01%     
==========================================
  Files         189      189              
  Lines       52588    52599      +11     
==========================================
+ Hits        52466    52476      +10     
- Misses        122      123       +1     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[brian-pane]

I had some extra time today, so I updated this to address the remaining ec25519-related ctgrind errors.

[brian-pane]

(I just rebased this to make sure it works cleanly atop PR #174.)

[ctz]

Hi, thanks for working on this.

I think it should be done in a slightly different way -- low::ct calls should be done in library code, rather than tests. The reason for that is that the library code is the part that should have and express an accurate handling of which quantities are secret or public.

I think that means that the public API has to change, eg Ed25519SigningKey::as_seed() (and any others that return a reference to secret storage). I think copying 32 bytes not worth worrying about.

[brian-pane]

Thanks, that makes sense; I'll rework this to have Ed25519SigningKey::as_seed() etc return low::ct::public'ed copies.

[brian-pane]

@ctz I think this is ready for review now. I moved the ct::public calls from the tests into the various library functions that return (copies of) secrets.

[ctz]

Thanks!